77?g International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B4. Beijing 2008
143
The semantic description model is built in abstract level and
concrete level, which is showed in Figure 2 in Unified
Modeling Language (UML). It is an integrated model
consisting of three parts: Data Profile, Data Content and Data
Binding. Data Profile tells users what the data are about by
describing the profile of the geographic information, such as
the dataset name, data format, acquisition methods, involved
processing, spatial extent, spatial resolution, provider, theme
and so on. Data Profile belongs to abstract level description. It
is biiilt by introducing metadata ontology. Data Content tells
users what the data contain by describing the content semantics
of geographic information data in concrete level.
Conventionally, geometries of spatial data are explicitly
represented. But spatial relationships are not explicitly or
formally represented. In this paper, geometric ontology and
spatial relation ontology are introduced into Data Content to
explicitly describe spatial semantics of geographic data. In
addition, fundamental geographic information ontology is
introduced to describe non-spatial data semantics. Data
Binding tells users where and how to access the data by
providing URL, contact information and procedure to access
data.
1
Figure 2. Ontology-based semantic description model
4. DISCOVERY AND RETRIEVAL BASED ON
ONTOLOGIES
Conventionally, discovery and retrieval for geographic
information is carried through based on keywords. However,
keywords are not sufficient to find exactly suitable geographic
information because lack of semantics and inference
mechanism usually causes search results often to be too narrow
or too large (Hartwig H. Hochmair, 2005). The emergence of
ontology provides possibility to enhance discovery and
retrieval. In this section, we discuss an ontology-based
approach for discovery and retrieval of geographic information.
4.1 Conceptual framework
The key to improve discovery and retrieval is to solve
problems of semantic heterogeneity between user’s search and
description of geographic information in SDI. Thus, user
ontology is needed to cooperate with ontologied introduced
into the semantic description model proposed above. User
ontology models concepts and relationships between concepts
in user’s view during search.
Figure 3. The hybrid ontology approach (Visser, U. & H.
Stuckenschmidt, 2002) (modified)